Skip to main content
SpringerLink
Log in

Eigenvalues Driven Gaussian Selection in continuous speech recognition using HMMs with full covariance matrices

  • Published:
Applied Intelligence Aims and scope Submit manuscript

Abstract

In this paper a novel algorithm for Gaussian Selection (GS) of mixtures used in a continuous speech recognition system is presented. The system is based on hidden Markov models (HMM), using Gaussian mixtures with full covariance matrices as output distributions. The purpose of Gaussian selection is to increase the speed of a speech recognition system, without degrading the recognition accuracy. The basic idea is to form hyper-mixtures by clustering close mixtures into a single group by means of Vector Quantization (VQ) and assigning it unique Gaussian parameters for estimation. In the decoding process only those hyper-mixtures which are above a designated threshold are selected, and only mixtures belonging to them are evaluated, improving computational efficiency. There is no problem with the clustering and evaluation if overlaps between the mixtures are small, and their variances are of the same range. However, in real case, there are numerous models which do not fit this profile. A Gaussian selection scheme proposed in this paper addresses this problem. For that purpose, beside the clustering algorithm, it also incorporates an algorithm for mixture grouping. The particular mixture is assigned to a group from the predefined set of groups, based on a value aggregated from eigenvalues of the covariance matrix of that mixture using Ordered Weighted Averaging operators (OWA). After the grouping of mixtures is carried out, Gaussian mixture clustering is performed on each group separately.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

References

  1. Bahl LR, de Souza PV, Gopalakrishnan PS, Nahamoo D, Picheny MA (1991) Context dependent modeling of phones in continuous speech using decision trees. In: Proc DARPA speech and natural language processing workshop, Pacific Grove, pp 264–270

  2. Biggs MC (1975) Constrained minimization using recursive quadratic programming. In: Dixon LCW, Szergo GP (eds) Towards global optimization. North-Holland, Amsterdam, pp 341–349

    Google Scholar 

  3. Bocchieri E (1993) Vector quantization for efficient computation of continuous density likelihoods. In: Proc ICASSP, Minneapolis, MN, vol 2, pp II-692–II-695

  4. Coleman TF, Li Y (1996) An interior, trust region approach for nonlinear minimization subject to bounds. SIAM J Optim 6:418–445

    Article  MATH  MathSciNet  Google Scholar 

  5. Gales M (1999) Semi-tied covariance matrices for hidden Markov models. IEEE Trans Speech Audio Process 7(3):272–281

    Article  Google Scholar 

  6. Huang XD, Lee KF, Hon HW (1990) On semi-continuous hidden Markov modelling. In: Proc ICASSP, pp 689–692

  7. Hunt M, Lefebre C (1989) A comparison of several acoustic representations for speech recognition with degraded and undegraded speech. In: Proc ICASSP, pp 262–265

  8. Kannan A, Ostendorf M, Rohlicek JR (1994) Maximum likelihood clustering of Gaussians for speech recognition. IEEE Trans Speech Audio Process 2(3):453–455

    Article  Google Scholar 

  9. Kay SM (1993) Fundamentals of statistical signal processing: estimation theory. Prentice Hall, New York

    MATH  Google Scholar 

  10. Knill KM, Gales MJF, Young SJ (1996) Use of Gaussian selection in large vocabulary continuous speech recognition using HMMs. In: Proc int conf spoken language processing

  11. Lindo Y, Buzo A, Gray RM (1980) An algorithm for vector quantizer design. IEEE Trans Commun COMM 28:84–95

    Article  Google Scholar 

  12. O’Hagan M (1988) Aggregating template or rule antecedents in real time expert systems with fuzzy set logic. In: Proc of the 22-th annual IEEE Asilomar conferences on signals, systems and computers, Pacific Grove, pp 681–689

  13. Shinoda K, Lee C-H (2001) A structural Bayes approach to speaker adaptation. IEEE Trans Speech Audio Process 9(3):276–287

    Article  Google Scholar 

  14. Simonin J, Delphin L, Damnati G (1998) Gaussian density tree structure in a multi-Gaussian HMM based speech recognition system. In: 5th int conf on spoken language processing, Sidney, Australia, 4 December 1998

  15. Watanabe T, Shinoda K, Takagi K, Iso K (1995) High speed speech recognition using tree-structured probability density function. In: Proc int conf acoust speech signal process, vol 1, pp 556–559

  16. Webb A (1999) Statistical pattern recognition. Oxford University Press, London. Arnold a member of the Hodder Headline Group, 338 Euston Road, London NW1 3BH, Great Britain

    MATH  Google Scholar 

  17. Yager RR (1988) On ordered weighted averaging aggregation operators in multi-criteria decision making. IEEE Trans Syst Man Cybern 18:183–190

    Article  MATH  MathSciNet  Google Scholar 

  18. Yager RR, Kacprzyk J (1997) The ordered weighted averaging operators, theory and applications. Kluwer Academic, Dordrecht

    Google Scholar 

  19. Yager RR, Rybalov A (1996) Uniform aggregation operators. Fuzzy Sets Syst 80:111–120

    Article  MATH  MathSciNet  Google Scholar 

  20. Yager RR, Rybalov A (1998) Full reinforcement operators in aggregation techniques. IEEE Trans Syst Man Cybern 28:757–769

    Article  Google Scholar 

  21. Young SJ, Odell JJ, Woodland PC (1994) Tree-based state tying for high accuracy acoustic modeling. In: Proc of the workshop on human on human language technology, pp 307–312

Download references

Author information

Authors and Affiliations

  1. Faculty of Technical Sciences Novi Sad, University of Novi Sad, Novi Sad, Serbia

    Marko Janev, Niksa Jakovljevic & Vlado Delic

  2. Alfanum Speech Technologies, Novi Sad, Serbia

    Darko Pekar

Authors
  1. Marko Janev
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Darko Pekar
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Niksa Jakovljevic
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Vlado Delic
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Marko Janev.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Janev, M., Pekar, D., Jakovljevic, N. et al. Eigenvalues Driven Gaussian Selection in continuous speech recognition using HMMs with full covariance matrices. Appl Intell 33, 107–116 (2010). https://doi.org/10.1007/s10489-008-0152-9

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10489-008-0152-9

Keywords

Search

Navigation